Advances in Alzheimer's Detection: A Multi-Learning Fusion Approach Using Choroidal Neovascularization Analysis

Alzheimer's disease (AD) is one of the brain disorders diseases that have a significant impact on the daily life of an affected person. Recently, several studies have revealed that there is a strong connection between several retinal and choroidal pathologies, such as Choroidal Neovascularization (CNV), which leads to AD. This research mainly focused on detecting AD from CNV via optical coherence tomography (OCT) images. OCT is used primarily to diagnose retinal diseases, as it provides high-resolution images of retinal layers and highlights the abnormalities present in the retinal images, which leads to CNV. Linking the CNV with retinal features of Alzheimer's requires unique techniques and analysis, as these two brains and eyes are interconnected by sharing the vascular and neural systems. In this work, the pre-trained model ResNET-101 with transfer learning is used to find the abnormal patterns belonging to AD using OCT images. The proposed Multi-Learning Fusion Model (MLFM) combines residual layers with Xtensible Convolutional Neural Networks (X-CNNs), which detect abnormalities from OCT images. In this context, the training model's residual layers transform the features of vascular and neural systems. Furthermore, the MLFM exhibits high sensitivity in detecting subtle choroidal changes associated with AD progression. Finally, the quantitative results show that the proposed MLFM obtains an accuracy of 99.78% with accurate abnormal region detection.